SYNOPSIS An assessment of three simple methods of typing Proteus strains is described. The methods chosen were biochemical typing, bacteriocine typing, and typing by means of the Dienes phenomenon. Dienes typing was deemed to be superior to biochemical typing and bacteriocine typing.A brief discussion on the relationship between the Dienes phenomenon and bacteriocine production is appended.Proteus infections in hospital practice may present a therapeutic problem, since Proteus species readily develop resistance to antibiotics (Carpenter, 1964).The extent to which Proteus cross-infection occurs is hard to assess, and in evaluating this problem difficulty arises because the infection may be exogenous or endogenous; since it is impossible to distinguish between these two routes of infection this problem must remain unsolved until a suitably simple and accurate tyDing system is made available.Several methods have been used: serology (Belyavin, Miles, and Miles, 1951;Krikler, 1953; Nestoresco, Popovici, Racovitai, Dimitriu, Floresco, Mermezan, Filipesco, Thomas, and Dragan, 1963), biochemistry (Kippax, 1957;Huang, 1966), Dienes typing (Krikler, 1953;Story, 1954;De Louvois, 1969;Skirrow, 1969). Bacteriocine typing (Cradock-Watson, 1965) and bacteriophage typing in conjunction with biochemistry (France and Markham, 1966) have all been employed.We wished to select a method suitable for a diagnostic laboratory rather than a reference laboratory, and accordingly chose the simplest of the available methods for study. For this reason, serology and bacteriophage typing were excluded as being too cumbersome for such a purpose. The remaining methods-biochemistry Dienes typing, and bacteriophage typing-were selected for evaluation. It was thought that these methcds were particularly suitable, as they required only simpleequipment and were not time-consuming. 'Present address: